Lookalike search finds creators who resemble a set of . You provide one or more examples of what you’re looking for, and the API returns ranked results based on content themes, audience overlap, posting style, and engagement patterns.This is useful when you have creators that work and want to find more like them — or when you need to replace a creator who declined and want someone with a similar profile.
A seed is a reference creator. You can identify seeds two ways:
// By creator ID (if you already have one from search or a previous lookup){ "creator_id": "a1b2c3d4-e5f6-7890-abcd-ef1234567890", "weight": 1.0 }// By platform + username (if you're starting from a social handle){ "platform": "instagram", "username": "fitness_coach_jane", "weight": 1.0 }
You can mix both formats in the same request. If a platform + username can’t be found, the API skips that seed and continues with the rest — the request only fails if none of the seeds resolve.
A higher weight means the API leans more heavily on that seed’s characteristics when ranking results. This is where lookalike search becomes genuinely useful beyond simple “find similar” — you can encode your preferences into the weights.
If you have campaign performance data (ROI, conversion rates, engagement), turn it into weights. Your best-performing creator gets weight 1.0, and others scale relative to that.For example, if you ran a campaign with three creators:
Creator
Campaign ROI
Weight
@top_performer
4.2x
1.0
@decent_creator
2.1x
0.5
@niche_pick
1.3x
0.3
The resulting lookalike search finds creators who skew toward your best performer’s profile while still incorporating traits from the others. Over time, this creates a feedback loop — each campaign’s results improve your next lookalike search.
The similarity.score ranges from 0 to 1. Scores above 0.8 typically indicate strong overlap. The shared_traits list explains what the seed and result have in common — useful for displaying to end users or for your own review.
Start with one seed to calibrate. Add more seeds once you understand what the single-seed results look like.
Use weights to encode knowledge. If you know one creator drives better results than another, the weights should reflect that. Don’t default everything to 1.0 if you have data.
Combine with match scoring. Run lookalike to expand your list, then pass the results through POST /v1/creators/match with your campaign brief to filter for actual campaign fit. Lookalikes find similar creators — match scoring tells you if they’re a good fit for this specific campaign.
